CN1976629A

CN1976629A - Medical imaging system for accurately measuring changes of orientation tumour

Info

Publication number: CN1976629A
Application number: CNA2005800213223A
Authority: CN
Inventors: D·F·杨克洛维茨; A·P·里夫斯; C·I·亨施克
Original assignee: Individual
Current assignee: Individual
Priority date: 2004-04-26
Filing date: 2005-04-25
Publication date: 2007-06-06
Also published as: WO2005104943A2; AU2005237517B2; EP1744671B1; EP1744671A2; US20070100226A1; CA2564240C; AU2005237517A1; EP1744671A4; CA2564240A1; US9033576B2; US7876939B2; US20160015355A1; US20110116606A1; HK1103001A1; JP2007534447A; JP5036534B2; WO2005104943A3

Abstract

A body part (204) is scanned (20) to produce a first set of imaging data (214A). A target lesion (5, 202A) in the imaging data is identified (30). The body part (204) is rescanned (40) at a subsequent time so as to produce a second set of imaging data (214B). The target lesion (5A, 202B) is identified in the second set of imaging data and the size of the target lesion (5, 202A) is measured in the first and second sets of imaging data to determine two apparent image volumes corresponding to the first and second sets of to imaging data (60). A change in size is estimated (70) by comparing the first and second apparent lesion sizes (30lA, 301B). A variance on the change in size is estimated (80) so as to determine a bound on the change in size measurement.

Description

Be used for accurately measuring the Medical Image System that directed tumor changes

Technical field

The present invention relates to the medical image data analysis generally, automated computer method particularly, with accurate mensuration through the directed tumor (target lesion) of Medical Image System imaging or the variation of multiple directed tumor.

Background technology

Field of medicaments develop various need be through the product of FDA official approval, this approval often is based upon on the measurement result that draws from medical image.The drug development cost relates to clinical trial so that anticarcinogen such as anticarcinogen can be given the ratification with one side the most consuming time at most.This point is especially true for oncology, also is applicable to the other medicines field certainly.

In oncology, assess the reaction of anticarcinogen treatment very usual with medical image now.Many clinical trials use the measurement result that abnormality or tumor (as tumor) are changed dimensionally as the leading indicator of evaluating therapeutic effect.The main foothold when although the variation of patient's survival rate is considered to be in the evaluation effect of drugs, but as the means that obtain FDA approval, this criterion (requisite) and tumor size change its criterion of this replacement and compare and less be given consideration.For example, being used for the treatment of lung cancer drugs may evaluate according to tumor in the lung or this standard of other tumor size minimizing speed.

RECIST (the therapeutic evaluation standard of solid tumor) standard be a kind of formal, set up be used to measure the method that tumor size changes.RECIST comprises one group of disclosed rule, these rule definitions when the cancer patient situation belongs to improve (" responding "), when belongs to no change (" stablizing ") during treating, or when situation belongs to seriously (" deterioration ").This standard is open through international cooperation, and described cooperation comprises European cancer research and treated tissue (EORTC), American National cancer association (NCI) and Canadian national cancer association clinical trial group.(see Therasse, et al., " New Guidelines to Evaluate the Responseto Treatment in Solid Tumors; " Journal of the National Cancer Institute, Vol.92, No.3, Feb.2,2000,205-216.).At present, great majority evaluation solid tumor all is to use RECIST to the clinical trial of the objective reaction of treatment of cancer.

The key element of RECIST standard is to use single space measurement result, wherein selects to contain the image of tumor maximum cross section diameter, and obtains unidimensional maximum measurement result from this image.Then with this one-dimensional measurement result with compare at the similar image of the same tumor of another special time, thereby the assessment response situation.According to RECIST, complete reaction is defined as the disappearance of tumor, and partial reaction is defined as size and reduces 30%, and progress is defined as tumor size and increases 20%.RECIST does not detect the tumor of size less than 1cm.

When carrying out any measurement, accuracy is a very crucial problem.Unfortunately, be used to evaluate tumor at present the RECIST method of therapeutic response is very limited, because it does not consider the accuracy measured.As a result, whether this method need observe the variations that take place in a large number and treatment is responded to judge in one-dimensional measurement.This needs to so a large amount of variations make this method can not measure tumor size reliably.

In the former standard operation, the caliper measurement of being undertaken by the lonizing radiation worker also can be used for measuring tumor size.The accuracy of measuring is evaluated and tested by the measurement variability of skilled lonizing radiation worker when measurement model (phantom) or the actual tumor.The mistake relevant with the manual measurement length of tumor may be quite big.Similarly, owing to select similar image in the scanning that can not be reliably promptly separates in moment, therefore must rely on a large amount of variations, purpose is that this certainly variation is real, rather than a kind of measurement mistake.

Generally speaking, present method all do not provide have step of the present invention, accurately method for measuring, the inventive method uses volumetric method to come size up.Present method is measured the length of tumor on a tangent plane, and only measures on one or two direction, rather than measurement all three-dimensional elements relevant with tumor.

Summary of the invention

The invention provides the range of error that a kind of automatic mode is used for determining the change in volume measurement.The health part obtains the first group image data after scanning.Directed tumor in the identification image data.This body part is scanned once more subsequently and obtains the second group image data.Be identified in the directed tumor in these second group image data, and the size of measuring directed tumor in first group and the second group image data is to determine two apparent image volumes corresponding to the first and second group image data.Variation on the size is estimated by the size that contrasts described first and second groups of apparent tumors.Estimate the difference on the change in size, thereby determine the excursion in the dimensional measurement.

On the one hand, the invention provides a kind of shortening clinical trial time method, in the shorter time interval, learn the accurate method whether tumor responds by providing a kind of.

On the other hand, the invention provides a kind of method that can definitely measure the less degree variation of tumor.

Description of drawings

Although novel feature of the present invention elaborates in the claims, but the following detailed description of carrying out in conjunction with figure of the present invention's (no matter for structure or content) can more help to be appreciated and understood that the present invention, and other purpose of the present invention and feature, wherein:

Fig. 1 has shown the simplification block chart that is used for accurately measuring the system that assesses directed tumor variation, the image system imaging of this orientation tumor through making up according to an embodiment of the invention;

Fig. 2 is the high standard functionalization block chart according to the automatic mode that is used for definite radiographic measurement range of error of one embodiment of the invention structure;

Fig. 3 is the high standard functionalization block chart of another embodiment of method that is used for determining the radiographic measurement range of error of selecting according to of the present invention that embodiment makes up;

Fig. 4 schematic form has shown the CT image slice of whole lung (large pulmonary);

Fig. 5 schematic form has shown stacking image on the CT image and visual tumor boundary method;

Fig. 6 schematic form has shown stacking image on the CT image and the replacement scheme of visual tumor boundary method;

Fig. 7 A and Fig. 7 B schematic form have shown the stacking image on the CT image that obtains at different time and another selectable embodiment of visual tumor boundary method;

Fig. 8 A and Fig. 8 B schematic form have shown the stacking image on the CT image that obtains at different time and another selectable embodiment of visual tumor boundary method;

Fig. 9 A and Fig. 9 B schematic form have shown the stacking image on the CT image that obtains at different time and another embodiment of visual tumor boundary method;

Figure 10 schematic form has shown stacking image on the CT image and another embodiment selected of visual tumor boundary method;

Figure 11 schematic form has shown stacking image on the CT image and another embodiment selected of visual tumor boundary method.

The specific embodiment

At first, what should arouse attention is that although the present invention describes particular system and the method for analyzing medical image data in detail, as the radiological medicine data, this is not to limit, and just is used for illustrating that the present invention also can be used to analyze other categorical data.

The present invention is based upon on the forward position Knowledge Base of imaging technique, and these technology can scan tumor now, thereby can the whole gross tumor volume of imaging.In the past decade, obtained major progress by utilizing the 3D cubing to calculate the machine algorithm from the method for CT radiographic measurement tumor size.In addition, now image can isotropically obtain, and promptly resolution (resolution) almost is identical on x, y and z direction.Improved (advanced) image processing allows tumor and surrounding structure further to cut apart, and tumor boundaries is had better definition, thus the measurement result that is improved.

The present invention combines higher resolution imaging technique and improved formation method, thus comparison of tumor more accurately.In the method, can measure littler degree change, change still sure to measuring tumor size simultaneously.In addition, can measure variation, rather than use single one-dimensional measurement simply by measurement volumes.By this method, can carry out more complete assessment to data.

Measurement accuracy depends on many factors.By estimating the error relevant, then can judge the accuracy of any concrete measurement of tumor with these factors.As the result who understands measurement accuracy, can provide about tumor size change more among a small circle with the prediction severe disease.Therefore, can make a definite diagnosis severe disease as soon as possible, reliably by utilizing analysis of the accuracy, use be that tumor is littler, but change in size more accurately, rather than use existing RECIST standard.

An existing clinical testing data management system that is known as ELCAP management system (EMS) can be advantageously utilised in the method for the present invention.EMS provides all aspects of trial system, comprises the computer for analysis of long-range radio operator's identification and image data.The characteristics of EMS are can be more effective and managing clinical trials in time, thereby make the time of testing shorter, uses DATA REASONING and preferable quality control more accurately simultaneously.Owing to the amount that lacks the data that clinical trial feature that patient monitoring causes loses in test place also can improve by the web-based system that use has real-time feedback and a report.Except automatic mode, whole repeatability and accuracy will be improved with the semi-automatic method that area or cubing are limited in a kind of some descriptive border of lonizing radiation worker manual drawing that requires.

With reference now to Fig. 1,, it has shown the simplification block chart of the automated system that is used for the directed tumor variation of accurate mensuration, the image system imaging of directed tumor through making up according to an embodiment of the invention.Image system 2 is at different time t ₁And t ₂Produce image data.Directed tumor 5 in the image data that time t1 produces also appears in the image data subsequently, that produce during at time t2, is decided to be directed tumor 5A.Tumor (lesion) 5 and tumor 5A are same tumors, but think here treat with anticarcinogen for exemplary purposes after, directed tumor is different in the volume size of different time.This image data is handled through operation image processing software 7 in computer processor 6.Directed tumor can comprise tumor, tuberosity etc.Described image also can comprise calibration equipment 10, and this will further go through below.

Described Medical Image System 2 can advantageously comprise any known Medical Image System.Some useful known image systems comprise that computed tomographic scanner, nuclear magnetic resonance, positron emission image system, x-ray imaging system vascular get involved and angioscope/angiography step, ultrasonic image system and Medical Image System of equal value.Directed tumor 5 through scanning can advantageously comprise the tumor type that The World Health Organization (WHO) and RECIST standard spell out, and comprises mastoncus, lung tumor, melanoma, colon tumor, ovary and sarcoma.

In useful embodiment of the present invention, software 7 is operated automatically, accurately measures the size and the volume of directed tumor 5.By this method,, just can estimate the variation on the volume if obtain the difference on free between the image data of directed tumor 5.Method of the present invention has been measured the degree of measuring relevant error with each, and purpose is the estimation volume, and finally estimates the proportionality variation of volume.The automatic mode that carries out under computer control has accurate repeatability.The measurement error that is caused by pseudo-shadow has been estimated in bearing calibration.Model, simulation and real tuberosity (nodules) are used for characterizing its measurement accuracy according to different tuberositys and they in the corresponding outward appearance of image (as the CT image).

The various features of assessing tumor changes in the measurement based on the given volume on its apparent volume with decision.Measurement will change according to the difference of tumor to the signal of background.The measurement error variation can advantageously comprise the estimation to the various positions of tumor (as the tuberosity with particular edge feature).

Therefore, for the boundary definition that provides at particular edge, can estimate according to the variation of measuring.As described below, also can advantageously estimate other factors, comprise that degree and contiguous structure that adjacent structure is connected with directed tumor estimate issuable influence to volume.The feature of measurement device also can comprise the factor that influences error variation.Cubing also can be subjected to the inherent discrimination rate of image system itself and exist in the influence of the quantity of the noise in the image.

Free-air correction

The bearing calibration of standard comprises model scanning and the measurement of noise content, pseudo-shadow and image distortion.Model is the culture with known dimensions.Because being spaces, the physical characteristic of image system such as CT scanner, these factors rely on.That is to say that measurement error changes according to measuring the position and the position of health in scanner of carrying out in vivo.Present operation can't utilize these factors, but the traditional all distortion map (conservativeglobal distortion figure) that can use manufacturer to provide.

By to scale-model investigation,, can advantageously set up the figure that can characterize image distortion degree, pseudo-shadow and noise at all region of interest of human body then for certain given image system.In case set up, this figure can be used for judging measurement of tumor measurement error scope more accurately.

Error correction

Adopted certain algorithm to organize the definite position of junction point by the measurement of computer accurately that the CT image carries out tumor size to judge tumor and other.This algorithm can be handled many dissimilar tumors, and uses different strategies to solve different situations.The foundation of error estimation is on image form and the specific algorithm processing method basis at this image employing.An aspect of of the present present invention, the data base produces at each identifiable image distortion, and measurement error estimation statistics variation in this data base draws.

According to the present invention, the method that is used for systematic error estimation comprises that (a) is to the measurement of the CT image of calibrated model with (b) to the multiple scaaning result's of the actual tumor of patient measurement.In a useful embodiment, can obtain by short scanning at interval the measurement of the tumor of slow growth.

As another embodiment, can obtain the repetition image of tumor with very short interval scanning, and no matter the speed of growth how, thereby obtain being based upon error estimation on the constant substantially tumor basis.This repetition image can obtain in the cut sections for microscopic examination process, wherein just can obtain the multiple video of tumor in several seconds.In addition, when the people participates in this measuring process, because the variation that causes of people's factor or error can be by relating to model or obtaining for the observer tests (human observer trials) through the people of the tumor of multiple scanning.

Relevant error can be by estimating a group model multiple imaging of simulating this position with specific geometrical position (as, the tuberosity that links to each other with thoracic wall).Variation between artificial model's scanning result can be used for characterizing its range of error at the position of each concrete test.

By multiple measurement, can obtain the desirable features of beam scanner system variation parameter to the model image.For example, the scanner reconstitution properties for example, point spread function can be by accurately judging scale-model investigation and experimental analysis.But model data can not be simulated all positions, because some tuberosity has the subtle difference that is difficult to set up model on density.In this case, a lot of tuberositys such, that do not have obvious growth are carried out multiple scaaning and just can be used to create the tuberosity data base.A method of accomplishing this point is two results that the same tumor of contrast obtains in very short time interval interscan.Then, described tuberosity data base just can be used to measure two measurement variations between the scanning result, and purpose is the measurement error of the given kind of the different tuberositys of estimation.

Specific image problem can produce specific pseudo-shadow.For example, heart movement can cause the fluctuation on the z direction in the 3-dimensional image shape.Another example is that the skeleton at the top can produce excessive noise.Can identify above-mentioned and other special situation, and range of error can be estimated from the data base under the similar situation and obtains.

Error estimation when needing manual intervention

In the position of some difficult imaging, the lonizing radiation worker may handle nodule segmentation and intervene.Then, the further processing of being undertaken by computer algorithm can make the lonizing radiation worker to the difference unanimity between the judgement of twice scanning result.The intervention that the lonizing radiation worker is undertaken by the data base who sets up is in this case given the credit in the estimation of measuring variation especially.In case judged the source of all measurement error, just can calculate the overall measurement error.

With reference now to Fig. 2,,, the figure illustrates the height functionalization block chart of the automatic mode that is used for definite dimensional measurement variation error scope according to one embodiment of the invention.Be used to judge that the automatic mode of cubing variation error scope comprises the steps:

In step 20, scan body part with image system, obtain the first group image data;

In step 30, at least one directed tumor of identification in image data;

In step 40, rescan body part to obtain the second group image data;

In step 50, at least one directed tumor of identification in the second group image data;

In step 60, at least one directed tumor of measuring imaging in the first group image data and the second group image data is to judge corresponding to the first apparent directed tumor size of the first group image data with corresponding to the second apparent directed tumor size of the second group image data;

In step 70, estimate change in size by contrasting the first and second apparent tumor sizes; With in step 80, the variation of estimation on the change in size is to judge the scope that changes in the dimensional measurement.

In step 80, this step is estimated in the variation of change in size and can be advantageously comprised the result that many factors that influence measurement accuracy are assessed.Can adopt the statistical method of standard to estimate or judge radiographic measurement variation and other error measure of discussing in the present invention.Such technology comprises, as linear regression, random-effect model, or the like.

The factor that influences measurement accuracy comprises error main source such as tuberosity form, scanner parameter, patient's factor, algorithm and operator's factor.Many in them are interactional.For example, point spread function, the patient that the definition on tuberosity border is depended on tuberosity tissue, scanner moves and other factors.The estimation that error is changed is to utilize at the image model of error component and from the measurement to the measurement of image model and patient to obtain the parameter of these models, also by computer simulation.Same patient is carried out to also can be used for reducing error observing.

The example of nodule shape factor comprises:

A. Density Distribution feature, as

I. evenly or the variable distribution feature, and/or

Ii. solid tissue or intragranular structure's feature.

B. the tuberosity geometric characteristic as

I. sphere or complicated shape, its complexity can be estimated, for example as the ratio of surface area to volume, is standardized as sphere (=1),

Ii. the shape of multiple component,

Iii. chamber, and/or

Iv. similar (close to) rebuilds the tiny characteristics of resolution.

Whether c. surface character such as tuberosity coarse (that is, having complex surfaces) or smooth, wherein coarse surface means high average curvature.

The example of scanner parameter comprises:

A. rebuild resolution and further comprise slice thickness, overlapping and/or interior pixel (pixel) size of face,

B.X-ray energy (dosage): kVp and mAs,

C. reconstruction filter,

D. the rotating speed of portal frame (gantry),

E. workbench (pitch) speed,

F. the point spread function of spatial variations, and/or

G. proofread and correct.

The example of patient's factor comprise following these:

A. the position of scanning area in health,

B. the size of health

C. air-breathing degree,

D. the air-breathing motion bottom of lung (particularly),

E. muscle spasm by a small margin,

F. the pseudo-shadow of speckle (streaking) for example, is transferred in the lung top, and/or

The health status of g. adjacent with tuberosity lung tissue indicates whether have cicatrix, emphysema or other and healthy relevant disease.

Operator's factor is from the operator, and it helps out in the tuberosity measuring process.For example, operator manual change possibly brings measurable factor through the tuberosity scope of estimation to measurement error, and this point can characterize by observer's research.

Full automatic algorithm has and the approaching position of inherent commit point (decision points) usually.For example, automation algorithm may be with the lump around the tumor as being the blood vessel that links to each other or as being a nodular part.It is how near from commit point that algorithm can be designed to their operations of indication, thus the indication error component relevant with dropping on the commit point opposite side.

Can represent tuberosity in case certain imagery zone is determined, then the variation of Ce Lianging just can be estimated by consideration as following described image model factor:

1. density: low variation is relevant with even solid tissue Density Distribution.High variation and high image noise and low or the spatial variations Density Distribution is relevant.

2. shape: low variation is relevant with spherical form, and high variation is irregularly shaped relevant with the height that contains many lumps or chamber.

3. surface character: at nodular border (edge), low variation is relevant with high image gradient, and high variation is relevant with low image gradient.Lower variation is relevant with smooth surface, and high variation is relevant with the irregular surface with high curvature.Frontier district between tuberosity and other related entities structure such as blood vessel or the thoracic wall (here have considerably less or at all do not have the image gradient sign on border) must be handled with different modes.For low variation, these frontier districts should be complementary between two scanning results of image partitioning algorithm.Since these zones can not be judged accurately as the gradient edge that therefore non-gradient is directly relevant with variation with the ratio that the gradient edge surface amasss.Fig. 4-Figure 11 with reference to the back discusses the distribution on the border of carrying out according to the present invention and the merging (incorporation) of boundary accurate at this.

4. size: usually, tuberosity is big more, and the ratio of part voxel (voxels) is more little, and it is accurate more that volume is estimated.Low variation relevant with greater tubercle (or very trickle scanner discrimination rate), and greatly variation common with than lesser tubercle relevant (supposing analog structure complexity (shape)).

Situation about may use through the variation of estimation comprises:

A. when two scanning results when all being available, all image datas and parameter all are considered to provide the scope through the growth rate of estimation.

B. when to have only a scanning result be available, be used to judge wait and carry out the minimum time of scanning for the second time that purpose is the decision that obtains having clinical significance through the variation of estimation.It is the time of measuring malignancy speed in the measurement error scope.

In some cases, be on two dimension (2D) area of single image rather than the volume that obtains from a group image estimation measurement size.

In the preferred embodiment of the invention, each step all is to carry out through suitable software, and this software allows medical science staff's participation.One embodiment of the invention further are included in the step at least one directed tumor edge of definition in the image data.The edge definition can be by adopting threshold value and/or gradient function to judge the border at this edge at least one directed tumor.For further assisted diagnosis, the software of use has adopted known in the art cutting apart automatically and classification technique determine the to hang oneself border and the fragment feature of body part (as lung) of image system imaging, comprises unusual.

In another embodiment, method of the present invention comprises and estimates certain this step of ad hoc structure movement degree automatically.In another useful embodiment, method of the present invention comprises at ad hoc structure estimates this step of movement degree automatically, comprises the intensity of variation of measuring surface texture and directed tumor external structure.In lung, according to the position of directed tumor and the distance of heart distance, the variation of described movement degree is very obvious.

In another useful embodiment, the method for wood invention comprises automatic coupling this step of corresponding image at least one directed tumor of different time acquisition.For example, this software may be chosen in has maximum sized directed tumor in the image, and with itself and comparable directed tumor contrast in second, the image that next obtains.Dimensional measurement can advantageously comprise length, area and the three-D volumes of tumor.

In another useful embodiment, method of the present invention comprises at least one directed tumor of selecting to have maximum area as target, and finds a step in the comparable target of ensuing time acquisition.

In another useful embodiment, method of the present invention comprises the step of utilizing at least one model, noise-measuring, the pseudo-shadow of scanner and image distortion to come the free-air correction image system.

With reference now to Fig. 3,, the figure illustrates the height functionalization block chart of the method that is used for definite radiographic measurement range of error.According to one embodiment of the invention, the step of this method comprises:

In step 120,, produce group image data with image decorum scanning body part;

In step 130, measure at least one the directed tumor that is imaged in these group image data, thereby judge apparent directed tumor size corresponding to these group image data;

In step 140, at least one error of estimating first apparent directed tumor size changes, thereby the estimation of whole measurement accuracy is judged;

In step 150, utilize the estimation of described whole measurement accuracy to judge the scope of directed tumor size; With

In step 160, judge that on the basis of the estimation of whole measurement accuracy time range to carry out measurement second time, shows clinical change.

This method that relates on the one hand of the present invention is preferably finished by the software that use is installed on the personal computer.In the preferred embodiment of the invention, the change in size of the directed tumor of indication severe disease is littler than RECIST standard code.The step of estimating at least one error parameter advantageously comprises (a) and calculates the radiographic measurement error of the calibrated model that obtains by CAT scanner and (b) calculate measurement error to patient's tumor multiple scaaning result.

In another useful embodiment of the present invention, the software that is adopted further comprise handler module to obtain because the variation that people's participation causes, people model or known dimensions, that carry out through the tumor that scans repeatedly that use by oneself tests as the observer data of use.

In one embodiment, described that grouping error factor comprises that at least one is selected from following factor:

The point spread function of image documentation equipment and relevant reconstruction filter;

The scanner parameter;

By the pseudo-shadow that causes with the high-density objects of tuberosity on same image plane;

Patient's motion;

The variation of patient posture between twice scanning;

When the scanning lung, the variation of health or inspiration capacity;

Nodular size;

What link to each other with tuberosity causes the structure of obscuring easily;

The tuberosity variable density;

Scanner calibration;

Definition to the tuberosity border; With

When skilled worker author participates in operator's variation that measuring process is brought.

Described scanner point spread function can be estimated by the one group of test scan result who uses calibration model to obtain.This scanner point spread function also can utilize patient's 3D calibration model by scanning. because moulded dimension is known,, scanning is used to estimate because the information of any deviation that scanning work person parameter causes so providing.This deviation information can be used to image data to dwindle the error that causes owing to scanning work person parameter then.

Can measure different scanning worker parameter between at least two scanning results by a group model scanning result that has used two parameter settings, thereby estimation is because the volume deviation that the parameter difference causes.Ideal operation is to use two scanning results with identical parameters.

Pseudo-shadow can advantageously characterize by calculating the image noise figure of forming based on the spatial frequency in interested target site, for as tuberosity or tumor.The data that pseudo-shadow also can obtain by the concordance research of carrying out with model characterize, and described model has similar noise figure, and other parameter provides the estimation to variation.

Patient is in the motion could affect result of scan period.The common form of patient motion comprise heart movement, patient's muscle spasm, breathing, pulse vibration or other form in directed tumor (as tuberosity) patient's motion of scan period.For example, through heart beat patient's kinematic error of characterizing be by nodule surface in imaging on the multiple variation of z direction of principal axis detect.Except patient moved, patient posture also influenced the image result.The variation of patient posture is to measure by the compatibility of 3D rigid body (rigidbody) position of contrast between at least twice scanning that different time carries out in twice scanning.

The variation of patient's state of an illness can be measured by the coupling of the 3D between twice scanning result.Breathing error change can estimate the research of tables of data scanning by utilizing.Go up big variation for breathing, can be used to estimate deviation and the variation that this situation causes the research that scans tables of data.

When directed tumor was tuberosity, the tuberosity scale error can characterize by the scale-model investigation of having used different size usually, to judge the inherent measurement error of the tuberosity size that certain is given.Similarly, the error that is caused by connected structure can characterize by carrying out model data and the measurement variation of connected structure under different condition that multiple scaaning obtains.Connected structure comprises the adnexa as organ junctional complex or density analogous organs.The error that is caused by connected structure can advantageously characterize by the nodular data from known dimensions, and this tuberosity has to be used for contrasting cuts apart conforming multiple scaaning result and adnexa.

Because the scanning work person proofreaies and correct the error that causes and also can characterize by the rectangular histogram adjustment method of using the image noise in the local image statistics.Because the error that scanner calibration causes also can characterize by using the calibration model with body part scanning.

In one embodiment, the error that is caused by the tuberosity boundary definition can characterize by contrast tuberosity boundary graph and point spread function.The error that is caused by the tuberosity boundary definition also can characterize by carrying out scale-model investigation, to judge the variation in the estimation of different condition lower volume.The error that causes by tuberosity density can be by the contrast known dimensions the multiple scaaning result of the tumor of growing slowly characterize.

In another embodiment, can measure by people observer research because the operator changes the error that causes, this research needs many lonizing radiation workers to participate in, and estimates the variation of these people under the different image quality conditions.

As mentioned above, multiple source of error is arranged when measuring.Select certain operator scheme when scanning, slice thickness is constant can control some error component as keeping.Other factors is the scanner intrinsic factor, as the modulation transfer function (MTF) of scanning system.In some cases, such intrinsic factor such as MTF may be scanned system manufacturer and point out in detail.At present, also not at the imaging side that is used for the cancer measurement of correlation and generally recognized standard.Yet error component is enough to improve a confidence level given, that the use error variation is measured to the influence of measurement accuracy, and measurement accuracy measure can be according to the present invention the estimating or derive of discussion.The other method that obtains the higher credibility of measurement accuracy is all to scan patient with calibration equipment each time.

Refer again to Fig. 1, no matter when patient is scanned, and in the time will carrying out the volume estimation, the present invention randomly comprises use calibration equipment 10.Described calibration equipment can comprise manikin, and this model also is scanned in patient's scanning.By this method, manikin will stand same sweep parameter with patient.This calibration equipment can be placed on the scanning center place, and/or, in addition, can give patient with calibration equipment, allow them carry this equipment.This calibration equipment advantageously contains the manikin of one group of different size.Described manikin can comprise one group of spherical object and one group of complicated structure through height correction.

In one embodiment, described calibration equipment can be placed in acrylic acid or the plastic castings, can be quite little.For example, size range being easy to carry and being equivalent to shandardized envelope, common book or the equipment of similar article size and can using for about 2cm * 2cm * 2cm according to required size.Under some scanning situation, big or also be suitable than skinny device.Other calibration equipment can comprise electric wire, pearl, bar or the similar articles of known dimensions and/or density.This equipment can be placed on patient and stand identical sweep parameter on one's body when scanning.Object in the measurement model then.Multiple object (on size and density through the height correction) measurement that can make a variation of using different sizes and type is to be used to consider deviation and repeatability.In this method, when given patient is scanned, can the measurement accuracy of the given scanner of estimation be set by using specific instrument.By using other known information that measurement accuracy is further enhanced about this scanning device, as above intrinsic factor of discussing such as MTF.

The embodiment of a replacement of the inventive method can be used calibration equipment or device in the body.For example, the electric wire of known dimensions, pearl, conduit, implantable devices or similar article can be placed on and be used in the patient body proofreading and correct or other medical application.Interior equipment of this body or element can be used for the correct scan result and will connect in scanning result and the error that different time obtains, and will connect between the different scanning situation, or either way comprise.

With reference now to Fig. 4,, this figure is by big lung tuberosity CT image slice figure.CT image 214 shows lung tuberosity 202, and this tuberosity contains the piece that connects together substantially in 208 districts of pulmonary 204.Other physical trait comprises spinal cord part 206 and other feature that links to each other with lung 210 and 212.Lung tuberosity 202 typically comprises the spicule that sends from tuberosity, and Fig. 6 has clearly shown this point.

Those skilled in the art understand the often unintelligible demonstration tumor of typical C T image (as tuberosity) and the definition border of feature on every side.

With reference now to Fig. 5,, tuberosity border visual method is presented on the CT image of overlapping imaging substantially.In preferred embodiments, different dotted lines 218,220 and 222 has been represented the border of coloud coding, and explanation is the source of error district.In one embodiment, dotted line 220 may be corresponding to the light green color border, illustrate be here good definition the tuberosity limit (as, have height image gradient); Therefore, the expectation error on green border is just very little.Dotted line 222 illustrates that corresponding to the light red border this regional image gradient is low, illustrates that perhaps this zone has the very small feature (being called spicule) of tumor, therefore can ignore from the volume estimation.The accuracy that the existence of low image gradient or pin dress thing reduces to measure.Dotted line 218 is corresponding to light blue border, illustrates that this zone has considerably less or do not have the image gradient evidence on border.In this case, the lonizing radiation worker can be allowed to use interactive software to come the position on artificial judgment border.Zone with low image gradient can be the source that produces maximum boundary alignment error.Adopt in such a way, the placement accuracy on tuberosity border is by visual, thereby indicates source of error and the possible size of error.

In one embodiment of the invention, the border of coloud coding can be plotted on the display in conjunction with information disclosed by the invention automatically by known graphics software technology.For example, can on the basis of the sum of errors correlated error variation relevant or other parameter (determining), select color according to above-mentioned disclosed content with given border (determining) by searching edge software.Can provide suitable key word or explanation and explain result or the image that demonstrates to help the operator.

With reference now to Fig. 6,, it is another replaceable embodiment of tuberosity border visual method, the overlapping demonstration in the CT image in schematic form of this scheme.In this interchangeable embodiment, tuberosity can advantageously comprise painted boundary line, for example comprises yellow 224, light blue 218, light green color 220 and light red 222, and the color is here represented by dissimilar dotted lines.The bilateral boundary line of enclosing region 208 can be used to the estimation scope of specification error.That is, real tuberosity border is considered to be in the bilateral boundary line.In this embodiment, yellow 224 are used to delineate trickle feature, and as spicule, these trickle features are considered to nodular composition, but when the measurement processing relates to the tuberosity volume it have been ignored, because they also are the very big sources of measurement error.

Such spicule comprises complex (but not comprising medically inessential structure), and according to the inventive method, these complexs are learned processing by statistics, as outlier.Such structure is normally long and thin, but volume is very little.Usually, ignore the small size structure relevant, thereby can not distort the measurement accuracy result with height error.

With reference now to Fig. 7 A and Fig. 7 B,, it is another replaceable embodiment of tuberosity border visual method, and this scheme is overlapping in schematic form to be presented in the CT image that different time obtains.Here, use and the above-mentioned similar scheme of mentioning of visual scheme, wherein, can obtain at least two scanning results of tumor, and wherein the difference between the scanning result also can be by visual.

Fig. 7 A has shown first CT image 214A of the tuberosity 202A that obtains for the first time, and Fig. 7 B has shown second CT image 214B of the same tuberosity 202B that obtains for the second time.Border 218A, 220A and the 222A of coloud coding are applied among the first image 214A, adopt above-mentioned then with reference to figure 5 and 6 technology of

describing.Border

218B, 220B and the 222B of coloud coding are applied among the second image 214B, adopt above-mentioned then with reference to figure 5 and 6 technology of describing.

With reference now to Fig. 8 A and Fig. 8 B,, it is another replaceable embodiment of tuberosity border visual method, and this scheme is overlapping in schematic form to be presented in the CT image that different time obtains, and wherein nodular growth or other change in size are visual.Here, border 218A, 220A and the 222A with first image overlaps on border 218B, the 220B and 222B of the acquisition of second image.The coverage diagram of gained is revealed (being presented on computer monitor or other the suitable display), thereby provides the visuality of tuberosity size to change, and to the explanation corresponding to the measurement accuracy on colour barcode border.

Have benefited from it will be understood by those skilled in the art that of disclosure of invention bound technique disclosed by the invention and be not limited to embodiment.These method for visualizing have many possible variations, comprising:

1. the nodular 3 dimension perspective views that will derive from all image slice add labelling,

2. use transparent (as, painted) labelling so that the structure below it also can be observed,

3. use the dark line labelling,

4. use the labelling of calibrate, distance can be observed quantitatively,

5. add range scale and textual description, thereby embody quantitative measurement, and/or

6. the combination in any of above-mentioned variation.

With reference now to Fig. 9 A and Fig. 9 B,, it is another replaceable embodiment of tuberosity border visual method, and this scheme is overlapping in schematic form to be presented in the CT image that different time obtains, and wherein nodular growth or other change in size are visual.Here, cross-hauling district 301A and 301B are the zones that the tuberosity size changes between two CT image 214A and 214B.For example, this cross-hauling district can advantageously for example be presented on the color video monitor with cerise.Other color also can be used.

With reference now to Figure 10,, it is another replaceable embodiment of tuberosity border visual method, and this scheme is overlapping in schematic form to be presented in the CT image of having represented another embodiment, and wherein nodular growth or other change in size are visual.Here, cross-hauling district 303,305,307 and 309 can show with shades of colour, illustrates variation has taken place, and with which type of credibility change.In one embodiment, zone 303 is corresponding to yellow region, and this district has represented the variation with the relevant higher degree of initial tumor size estimation.Zone 307 relates to and the uncertainty of measuring relevant size for the second time corresponding to the green district.On behalf of this district, zone 307 have the height change probability corresponding to red color area.Zone 309 is corresponding to the blue area, and this district is illustrated in the zone that some measurement can overlap.This figure has also set a model, is used to inform how we operate in the time will measuring response.In addition, can advantageously select central point 320, thereby can change estimation various quadrants (quadrants) 320A, 320B, 320C and the 320D of piece.In some volumes, described variation may change much bigger than other, and certainty factor is also different.

With reference now to Figure 11,, it is another replaceable embodiment of tuberosity border visual method, and this scheme is overlapping in schematic form to be presented in the CT image, has equally also represented another embodiment, and wherein nodular growth or other change in size are visual.Figure 11 is basic identical with Figure 10, has also added boundary line 313 in addition, and this line is with wherein variation can not be by nodular that part of the putting together of reliable measurements (but variation useless can be judged reliably).This makes other position of tuberosity can access analysis.

Although the present invention has explained and described specific embodiments of the present invention, to those skilled in the art, can carry out many modifications and variations to the present invention.Therefore, it should be understood that the claim of enclosing is used to protect all such modifications and variations, they drop in theme of the present invention and the protection domain.

Claims

1. automatic mode that is used for determining change in size measurement error scope, the step that this method comprises is:

With image system (2) scanning (20) body part, obtain the first group image data;

Identification (30) at least one directed tumor (5) in image data;

Rescan (40) body part to obtain the second group image data;

Identification (50) at least one directed tumor in the second group image data;

At least one directed tumor (5) of measuring (60) imaging in the first group image data and the second group image data is to judge corresponding to the first apparent directed tumor size of the first group image data with corresponding to the second apparent directed tumor size of the second group image data;

Estimate change in size (70) by contrasting the first and second apparent tumor sizes; With

Variation (80) on the estimation change in size is to judge the scope that changes in the dimensional measurement.

2. method according to claim 1, wherein this dimensional measurement comprises and is selected from least one following dimensional measurement: the length of tumor, area or three-D volumes.

3. method according to claim 1, this method further comprises the step that defines the edge of at least one directed tumor (5) by the apparent image volume according at least one directed tumor of radiographic measurement variation adjustment, to produce at least two image volumes through adjusting.

4. method according to claim 3, the step that wherein defines the edge of at least one directed tumor (5) further comprise the step that threshold value and/or gradient function is applied to this at least one directed tumor, to determine the border at edge.

5. method according to claim 1, the step of wherein estimating radiographic measurement variation (80) further comprises the big measure feature of measuring at least one directed tumor (5), the feature of measuring adjacent structure, the step of image system feature, and described image system feature comprises the amount of the noise that exists in its inherent discrimination rate and the image.

6. method according to claim 1, wherein each step all is by using the appropriate software (7) that allows the medical worker to participate in to carry out.

7. method according to claim 1, this method further comprises the step of estimating movement degree at ad hoc structure automatically.

8. method according to claim 6, wherein the step of estimating movement degree automatically at ad hoc structure comprises the intensity of variation of measuring the outer structure of surface texture and directed tumor.

9. method according to claim 1, this method further comprise mates the step of at least one directed tumor at the corresponding image of different time acquisition automatically.

10. method according to claim 9, this method further comprise at least one directed tumor of selecting to have maximum area, greatest length or maximum volume as target, and find the step of the comparable target that obtains in the ensuing time.

11. method according to claim 1, this method further comprise the step with at least one model (10) free-air correction image system and measurement noise amount, pseudo-shadow and image distortion.

12. method according to claim 11, wherein this step of free-air correction further comprises for given scanner and carries out this step of scale-model investigation set up to characterize the figure of noise, pseudo-shadow and image distortion degree at the human body relevant range; And utilize this figure to determine the measurement error scope of the measurement of directed tumor (5).

13. according to the described method of claim 1, wherein body part is lung (204), this method further comprises the step that lung further feature and at least one directed tumor are cut apart automatically.

14. method according to claim 1, wherein said image system are selected from computed tomographic scanner, nuclear magnetic resonance, positron emission image system, x-ray imaging system, vascular intervention and angioscope/angiography step or ultrasonic image system.

15. a method of measuring the radiographic measurement range of error, this method comprises the steps:

Produce group image data (120) with image system scanning body part;

At least one directed tumor of measurement imaging in these group image data is to determine the apparent directed tumor size (130) corresponding to these group image data;

At least one error variation of estimating this apparent directed tumor size is to determine the estimation (140) to the overall measurement accuracy;

Use this estimation to determine the scope (150) of targeting tumor to the overall measurement accuracy; With

Determine the time frame in the estimation of overall measurement accuracy to be used to indicate the ensuing measurement (160) of clinical change based on this.

16. little than RECIST standard code of method according to claim 15, the change in size of at least one directed tumor that critical event (significant event) wherein is described.

17. method according to claim 15, the step of wherein estimating at least one error variation comprises (140) following step: (a) calculate measurement error and (b) calculate measurement error from the multiple scaaning result of patient's tumor from the computed tomographic scanner image of calibrated model.

The error variation step that the people who tests by the people observer who utilizes model or the tumor through scanning repeatedly 18. method according to claim 15, this method further comprise participates in obtaining.

19. method according to claim 15, the step of wherein estimating at least one error variation (140) comprises estimation owing to be selected from the error effect that following at least one factor is brought:

The scanner parameter;

Patient's motion;

The variation of patient posture between twice scanning;

When the scanning lung, the variation of health or inspiration capacity;

Nodular size;

The tuberosity variable density;

Scanner calibration;

Definition to the tuberosity border; With

20. method according to claim 19, wherein the point spread function of scanner is to estimate by the one group of test scan result who obtains with calibration model.

21. method according to claim 19, wherein said scanner point spread function are to utilize patient's 3D calibration model (10) to estimate by scanning.

22. method according to claim 19, wherein the group model scanning result that has been to use two parameters to be provided with of the different scanning worker parameter between at least two scanning results is measured, thereby estimation is because the volume deviation that the parameter difference causes.

23. method according to claim 19, wherein the threedimensional model scanning of carrying out with patient provides the information of estimation deviation, thereby reduces because the error that scanning person's parameter causes.

23. method according to claim 19, wherein pseudo-shadow characterizes by calculating the image noise figure of forming based on the tuberal region spatial frequency.

24. method according to claim 19, wherein said pseudo-shadow characterizes by the concordance research that use has the model (10) of similar noise figure, and other parameter provides the estimation to variation.

25. method according to claim 19, wherein patient's kinematic error is characterized by heart movement or pulse vibration, and this heart movement or pulse vibration are by the multiple change detection of z direction of principal axis on the imaging nodule surface.

26. method according to claim 19, wherein patient's motion is selected from tuberosity scan period heart movement or patient's muscle spasm.

27. method according to claim 19, wherein the variation of patient posture is to measure by the matching that contrasts the 3D rigid body position between this at least twice scanning in twice scanning.

28. method according to claim 19, wherein breathing error change is by utilizing scanning to be estimated the research of tables of data.

29. method according to claim 19, wherein the tuberosity scale error is to characterize by the scale-model investigation of using different size, measures with the inherence of judging the tuberosity size that certain is given to change.

30. method according to claim 19 is wherein owing to the error that connected structure causes characterizes by using the model data of carrying out multiple scaaning and measuring the connected structure of variation under different condition.

31. method according to claim 19, wherein since the error that causes of connected structure be the nodular truthful data by having adnexa and be used to contrast and cut apart conforming multiple scaaning result and characterize.

32. method according to claim 19, wherein the error that causes of tuberosity density characterizes by contrast multiple scaaning result.

33. method according to claim 19 is wherein characterized by using twice research block diagram coupling by the error that scanner calibration causes.

34. method according to claim 19, wherein the error that causes of scanner calibration is that calibration model when using scanning characterizes.

35. method according to claim 19 is wherein characterized by contrast tuberosity boundary graph and point spread function by the error that the tuberosity border causes.

36. method according to claim 19 is wherein characterized by carrying out scale-model investigation with model by the error that the tuberosity boundary definition causes, to judge the variation in the estimation of different condition lower volume.

37. method according to claim 1, the step that wherein scans body part further comprises scanning calibration equipment (10).

38. method according to claim 19, the step that wherein scans body part further comprises scanning calibration equipment (10).

39. an automatic mode that is used for determining change in size measurement error scope, the step that this method comprises is:

With image system (2) scanning body part (204), obtain the first group image data (214A);

Be identified in the most of directed tumor (5) of (214A) in the image data;

Rescan body part (204) to obtain the second group image data (214B);

Be identified in the most of directed tumor among the second group image data C;

The most of directed tumor (5) of measurement imaging in the first group image data (214A) and the second group image data (214B) is to judge corresponding to the first apparent directed tumor score value of the first group image data with corresponding to the second apparent directed tumor score value of the second group image data;

Estimate the variation (70) of directed tumor score value by contrasting the first and second apparent tumor score values; With

The variation of estimating in the directed tumor score value variation is measured the scope (80) that changes to judge score value.

40. method according to claim 15, wherein this at least one directed tumor (5) is less than 1cm.

41. the tuberosity border method for visualizing of an overlapping imaging on the CT image, the step that this method comprises is:

Obtain first image (214A) of tuberosity (208A);

Determine nodular first group of border (218A, 220A, 222A) corresponding to image gradient level;

With show first image, cover on first image (214A) border (218A, 220A, 222A) is overlapping simultaneously.

42. according to the described method of claim 41, this method further comprises at border (218,220,222,224) uses the step of different colours to indicate boundary error value scope.

43. according to the described method of claim 41, this method further comprises the step that spicule information (224) is excluded from measurement accuracy is calculated.

44. according to the described method of claim 41, this method further comprises the step of using interactive software to come the position on the selected border of artificial judgment.

45. according to the described method of claim 41, this method further comprises the steps:

Obtain second image (214B) of tuberosity (208B);

Second group of border (218B, 220B, 222B) is useful on second image (214B); With

On the display first group of border (218A, 220A, 222A) and second group of border (218B, 220B, 222B) are being covered to show any variation on the tuberosity size.

46. according to the described method of claim 45, the zone that wherein shows the tuberosity change in size is to show visually.

47. according to the described method of claim 41, wherein said first group of border is to utilize to be selected from following method for visualizing demonstration:

The nodular 3 dimension perspective views that derive from all image slice are added labelling,

Use transparent labelling, so that the structure below it also can be observed,

Use the dark line labelling,

Use the labelling of calibrate, so that distance can be observed quantitatively,

Add range scale and textual description, thereby show quantitative measurement,

Combination in any with said method.

48. according to the described method of claim 41, wherein the zone on image covers so that selected image factor (303,305,307 and 309) to be described with labelling.

49. according to the described method of claim 48, wherein selected image factor (303,305,307 and 309) is selected from zone that the higher degree relevant with initial tumor size estimation change, with the uncertain relevant range of measuring for the second time relevant size, have the zone of height change probability and the zone that in some is measured, can overlap.

50. according to the described method of claim 48, this method further comprises selects central point 320 steps, thereby allows nodular various quadrants (320A, 320B, 320C and 320D) are changed estimation.

51. according to the described method of claim 48, this method further comprises border (313) with wherein change can not be by reliable measurements, rather than the tuberosity part that can be judged reliably of the variation useless step of putting together.

52. method according to claim 1, this method further are included in the step of scan period use calibration equipment (10).

53. according to the described method of claim 52, wherein this described calibration equipment comprises manikin.

54. according to the described method of claim 52, wherein this calibration equipment (10) comprises the manikin of one group of various sizes.

55. according to the described method of claim 52, wherein this calibration equipment (10) is selected from the article of electric wire, pearl, bar or geometry.

56. according to the described method of claim 52, wherein said calibration equipment (10) is an equipment in the body.

57. according to the described method of claim 56, in the wherein said body equipment be selected from electric wire, pearl, conduit, implantable devices or known dimensions at the intravital article of patient.